The system since 70s to this day is like :
Power input into lowest floor --> power company protections main for building --> power company protections for each user --> meter --> user panel with user's main breaker
The main power company potections use one-time cartridges with high interruption capacity - sand filled. They are slow-blow and rated so that the user's main breaker will allways trip before them. They are there in case there is a short circuit in the input to the user panel before the main breaker (damage to main cable) or in the meter, and to prevent users from swapping up their main breakers and overloading the grid
Officially they are supposed to be in a box with the power company's seal, the home owner is not allowed to access them
In reality, in most older (up to 80s) homes the box is not sealed, and the users do tamper with them occasionally. A common reason why they blow "and need tampering" is users swapping up the main breaker so it won't trip when they overload, or users connecting a circuit for a big load (most commonly water heaters) from before the main breaker
The user panel breakers are rewirable - You pull out the cartridge, connect a stretch of new fuse wire between the terminals, and plug it back in
In later years there appeared breakers that fit in the user fuse holders in place of the cartridge
I disassembled one of those replacement breakers - a BBC (the company is known as ABB nowadays) PicoSTOTZ - im not impressed. Contacts dont open very far apart, and the arc extinguishing measures are inferior to what is found in DIN rail breakers. It probably would work for the small short circuit currents in a house wired entirely with 1.5, but i would not trust this thing to open any significant short circuit, atleast without exploding while doing that..
I have somewhere a Soviet made version too, those are screw in shape and allmost fit an ordinary E27 socket, except the base contact is longer then in a lamp. The Soviet things are scary. The mechanics get stuck simply from pressing the reset/release buttons a few times, and definitely do if you press both buttons together, sometimes rendering the thing stuck for good in some middle position. The opening contact is a double contact type, so even after the bridgeing part is away from the main contacts, they are still next to each other. No arc extinguishing measures present at all. The trip mechanics are very rough and there are few things that can randomly fail and get stuck when its time to trip
In place of the coin hack were other variations :
In the rewirable fuses, users would connect a piece of Cu wire in the cartridge instead of the fuse wire, or would jumper around the fuse holder (add a wire connecting directly between the top and bottom terminals). I seen a 50s home in BeerSheva, the entrance floor panel there (multiple lighting circuits of the entrance and basement floors) looked like : Fuse holders of various sizes, in all of them the cartridges are missing, and there are bundles of 1.5mm^2 Cu wire stuck between the terminals around the cartridge
In the power company fuses, users would empty the cartridge (spill the sand out) and put in Cu wire, or wrap the blown cartridge as is with Al foil and screw it back in
With the Al foil, some peeps tried to use the Al coated paper from cigarette boxes... The metal in those things is very thin, and it does act as a fuse. Often setting fire to the paper while doing so
Lower floor of a 70s flats building (of a friend of mine). The panel contains the main power company fuses for the building and for the staircase/shelter, and a user panel for the staircase/shelter (that was worked on by some bright spark)
http://www.lighting-gallery.net/gallery/displayimage.php?album=2438&pos=43&pid=5984560s fuse panel (found on the internet). The breaker box is a later addition, originally the sleeved wires from below (from the meter) were going straight into the fuse panel. Labels below fuses are (starting from left) : "Water Heater" "Appliances" "Appliances" "Lights" "Lights"
http://oi58.tinypic.com/2ngqssj.jpgWell made RCDs normally last for decades - Many 70s RCDs are still around and are fine
I seen RCD failures and had the one in my panel fail too. The most common failure mode is random tripping for no reason, or being impossible to reset after it tripped once for a real reason. starts with trips once in a month, then becomes more common, then one day you simply cant reset it at all, even when it is out of the panel and not connected to anything - It would mechanically not click into the "on" position anymore
The "randomly tripping for no reason" thing is hard to troubleshoot, as the exact same behavior happens when the RCD is ok and is tripping for a reason. The way i troubleshooted it when mine started random tripping - First i measured many times the circuits on it for leaks. When i was not able to find anything, i replaced the RCD, and the tripping stopped
One common reason for RCD and any breaker failures is building debris that lands from the wall materials above the panel at time of construction or when messing with the wall and the panel is open. The dirt fallson the breakers and gets inside them into the mechanics, probably through the upper hole for the wire
Some years back i was called (as IT) to check for an office full of computers that blew all at once. I traced it down to a damaged 3 Phase RCD, that closed the Phase contacts but not the Neutral. The panel was full with phenomenal quantities of cement dust, some of which got inside the RCD and happened to stick on the Neutral switching contact pads, keeping them apart while the Phase contacts closed properly. This probably happened after the RCD tripped (for a reason), then a bit of dirt got to where it got, and somebody reset the RCD with the dirt inside
Code here requires RCD for all circuits at home including lighting. The most common setup is as Medved says, main current limiting breaker --> one RCD for all --> individual circuit breakers. Few peeps do install separate RCDs for separate circuits
Last night i had an event with the RCD tripping :
2AM everything is still fine, i switch off the computer, the light, and go to sleep. Morning the RCD is down. Something happened in the mid of the night..
The fault can be traced by measuring resistance L-PE and N-PE in each circuit (from the panel, with the power off). Measure DC voltage first (just in case, to not thunderbolt attack my multimeter with the charge from some Y capacitor somewhere, in Ohms measurement mode), then measure Ohms
L-PE in some circuits shows infinity, in some capacitance charging up and then infinity, and in some resistances. But it is possible that measuring like this is actually measuring a circuit like : L on the tested circuit - some load resistance (primary of a transformer or motor, ...) - N on the tested circuit - common N in panel - N on another circuit - Short is in the other circuit. So without separating the N's in the panel the results of this test are not much useable..
N-PE (all N are on the Neutral bar in the panel) shows single Ohms
Take out all the wires from N bar, measure each to PE. The offending circuit is one of the lighting circuits
Take apart the connections in the boxes this circuit goes to, measure each N to PE. the offender was a Fluorescent light in the bathroom. This also explains why L-PE showed nothing. The short was touching metal short N-PE. To the multimeter (measuring at very low voltage), the Fluorescent lamp and starter appear as open
The problem was, the metal edge of the back of the lantern was pressed hard into the N wire where it comes out of the wall, and it was punched through at the sharp point of the edge. It was installed there since about 1999..2000, The wire was probably pressed all along, but not punched through yet. I banged this light accidentally a few days before while replacing the shower curtain rod, and apparently this was the last push it took to punch through the isolation of the wire completely
I have a guess why it tripped the RCD last night and not allready back then :
Maybe the wire did not make very good contact in the point where it was punched through, and the small N-PE voltage present did not cause 30mA of leakage current
A momentary high load (fridge compressor starting ?) made the current there momentarily rise, possibly making some power dissipate there and improve the contact conductivity with each subsequent current pulse (think oxide layer on the lantern's metal breaking down ?)
After a few 10's times of such pushes, the contact eventually became good enough to conduct 30mA at the peak N-PE voltage, and that is when it tripped